Common-mode EI PS Chokes

[rdf]

Three recurrent threads got me thinking power supply filtering - the one over low DCR supplies, another between chokes in the +ve vs. -ve supply leg and the last about using mH common-mode chokes to filter PS RF. Why can't we have all three? On the face of it no reasons come to mind that >1H power supply chokes can't be wound common-mode on an EI core. Since the DC balance is perfect core size will be much smaller, bandwidth should be much wider than common single-ended chokes, DCR low and as a bonus both B+ and ground are filtered.

Typical common-mode chokes seem to stop at around ~.2H, the technology is targeted towards line filters and switching supplies. Does anyone know of a manufacturer selling or industry employing higher values? Perhaps a custom winder? Barring that, I have a pile of old Hammond 10H, ~200 ohm EI chokes willing to donate parts but can't find any references to winding CM chokes on EI cores.

At this stage the apparent potential for reduced cost and weight and much improved performance is very attractive. Any info or direction greatly appreciated.

[jayme]

Lundahl sells C-core chokes that can be wired as a common-mode choke. I've used Lundahl iron in my preamp, and found it to be excellent.

[ErikdeBest]

Wouldn't it be possible to use a trafo wtih two common windings (as a insulation tranny) as a common-mode choke? Looking at the symbol of a common-choke I don't see why not...also, I don't have a clue how much inductance they would have (probably two 115V windings would held higher inductance than two 6V windings, although the last could hand more current).

Well, at least my post helped me to stay tuned to this thread. I am curious...and willing to learn

Erik

[Gluca]

Would CM chokes be effective in reducing the ripple? In LL datasheets chokes are not used as CM chokes.

Ciao
Gianluca

[rdf]

Thanks jayme. I'm hoping for something 'off the shelf' first if at all possible but it's a good option.

Hi Erik, I got to thinking the same thing. The only question is whether primary and secondary are wound identically. One promising option is a control transformer, which to my knowledge is designed to work symmetrically in either direction, implying identical windings. Relatively inexpensive too.

[jayme]

I meant to specify that the lundahls are exactly what you were looking for: two windings on a single core, tho isolated so that you can wire it up in common-mode, one widing per leg.

However, there have been some posts on either this forum or audioasylum that the common core causes the common-mode rejection to be frequency-dependent, due to low-frequency coupling.

I tried the Lundahl wired in common-mode for the first choke in a CLCLC PSU. The overall B+ voltage went up...from 315V to 330V. I've been told that this is due to providing a better load on the mains transformer, which increases its efficiency. I also understand that it presents an easier swing for the rectifier.

The downside was that I could not ground the PSU at the first cap. For me, this produced some hum, so I switched back to a normally configured choke in the positive rail.

[tubelab.com]

One Electron makes chokes intended for parafeed applications they have two identical windings that can be used in parallel or series. The RC-1 has two 11 HY 115 mA windings. I don't have one here to test.

I got a similar idea a few years ago, but tried a power toroid with two 120 volt windings. I guess it wasn't meant for 450 VDC between the two windings because it fried rather quickly on power up. I gave up on that idea. I have several industrial control transformers lying around in sizes from 200 VA to 2KVA (big!). I will try one of them as soon as I can get some bench space. It is easy to try on my test amp, but my bench is tied up right now.

[ErikdeBest]

The only question is whether primary and secondary are wound identically.

I was thinking on trafo's with two primary windings of 115V. From what I understood those windings must be balanced, otherwise there are problems when connecting bot windings in parallel for operation on 115V. Those trafo's can be had quiet cheap from surplus...

I guess it wasn't meant for 450 VDC between the two windings because it fried rather quickly on power up.

That is a less motivating story...I did not think about that, as most toroids have a insulation that goes far above the B+ I generally use in my projects, but that seems to be the insulation between primary and secondary, not between both primaries... That brings me to another question...in a trafo with 2x 115V winding and 2x12V winding (just an example)...how can the insulation between the 115 windings and the 12V windings be 2kV and the insulation between both 115V windings be much less?

Erik

[Dave Cigna]

The real challenge is getting enough bandwidth. I don't know exactly, but I had always assumed that the common mode stuff that you want to filter out is high frequency. A few hundred kHz I imagine, but maybe somebody else can be more precise.

Anyway, something like an isolation transformer, even if both windings have the same number of turns is likely to have pitiful bandwidth. It depends on the winding geometry; a split bobbin design would be the worst, but even something with one winding on top of the other would probably drop into the 10's of mH range above a few kHz. Why bother?

In fact, any EI core becomes more or less ineffective above a few 10's of kHz. With a 1:1 transformer (an interstage for example) you can get extended bandwidth with bifilar winding. At high frequencies the core is pretty much out of the picture, but the two windings stay coupled because of the close proximity. With bifilar there is relatively large capacitance between windings. In the case of a 1:1 transformer it actually works to your advantage; it sort of acts as a 'hidden' coupling cap. In the case of a CM choke it works against you; the 'input' lead of one winding would be coupled capacitively to the 'output' lead of the other winding and vice-versa. The capacitance effectively short circuits the choke at high frequencies.

One thing that might work is a toroid with one winding laid against the core and the other laid right on top. But, they would need to be wound in opposite directions to avoid the capacitance issue of a bifilar wind. The difficulty there is getting a custom winder to make the thing. It's relatively easy to get almost anything you want on an EI core, but toroids take special machinery and I don't know of any toroid winding houses that do one-off custom stuff.

I suppose you might get away with an EI core if you layer wind. First a layer of winding A, then a layer of B, then another layer of A, then another B, .... The A's would all be connected in series as would the B's. The worst part is that the A's and B's would have to be wound in opposite directions. A VERY tedious job. I wouldn't want to do it!

Another alternative would be two separate HIGH BANDWITH chokes, one in each leg. That would provide both common mode and differential mode filtering. You'd lose the benefit of the flux cancellation of both windings on a single core. They'd both carry DC, so they would need to be gapped and they'd be big.

A couple hundred mH for $10 doesn't sound so bad anymore.

-- Dave

[tubelab.com]

My idea was to filter out audio frequency crud that comes in on the power lines. Try plugging an electric drill into the same outlet as your amp. listen to the sound that the amp makes when you pull the trigger. Ideally there should be none. People spend $200 on power cords that are supposed to help. If this transformer idea works to 5 khz then we can wind a CM choke on ferrite to kill the rest.

I would think that the toroid should have handled 450 VDC since it waws made for 230 VAC. It is possible that the toroid already had problems. It came out of a junk box. It smoked instantly.

More science is needed as soon as I can get this computer off of my workbench.